+1 for Sunday night. This also helps Filesystem and Hive implementation for FLIP-27 Source. And the implementation will block "Support Multiple Input for Blink Planner". Multiple input is very important for Batch performance.
Best, Jingsong On Tue, Oct 20, 2020 at 6:36 PM Yu Li <car...@gmail.com> wrote: > +1 for Sunday night. This also helps for more thorough testing of the > RocksDB version bumping up job [1]. > > Thanks. > > Best Regards, > Yu > > [1] https://issues.apache.org/jira/browse/FLINK-14482 > > On Tue, 20 Oct 2020 at 17:06, Robert Metzger <rmetz...@apache.org> wrote: > > > Thanks a lot. > > > > It seems that a few people are supportive of the idea of moving the > feature > > freeze to Friday. > > I would actually propose to make it *Sunday night* then. We'll then > create > > the first release candidate on Monday morning, November 2nd. > > > > > > On Mon, Oct 19, 2020 at 1:27 PM Danny Chan <yuzhao....@gmail.com> wrote: > > > > > Per FLIP-145, we have many runtime operators to implement and bridge it > > > with the planner. > > > > > > Best, > > > Danny Chan > > > 在 2020年10月19日 +0800 PM6:59，Robert Metzger <rmetz...@apache.org>，写道： > > > > Thank you for your responses so far. > > > > > > > > @Kurt, Jingsong, Danny: Which JIRAs/FLIPs are going to benefit from > the > > > > extension? > > > > > > > > On Mon, Oct 19, 2020 at 12:45 PM Aljoscha Krettek < > aljos...@apache.org > > > > > > > wrote: > > > > > > > > > @Robert Your (and Dian's) suggestions sound good to me! I like > > keeping > > > > > to master frozen for a while since it will prevent a lot of > duplicate > > > > > merging efforts. > > > > > > > > > > Regarding the date: I'm fine with the proposed date but I can also > > see > > > > > that extending it to the end of the week could be helpful. > > > > > > > > > > Aljoscha > > > > > > > > > > On 19.10.20 12:24, Danny Chan wrote: > > > > > > +1 for Kurt suggestion, there are many features for SQL yet, 2 > more > > > days > > > > > are valuable. > > > > > > > > > > > > Best, > > > > > > Danny Chan > > > > > > 在 2020年10月19日 +0800 PM6:22，Jingsong Li <jingsongl...@gmail.com > > >，写道： > > > > > > > Hi Robert, > > > > > > > > > > > > > > Thanks for your detailed explanation. > > > > > > > > > > > > > > At present, we are preparing or participating in Flink forward, > > so > > > +1 > > > > > for > > > > > > > appropriate extension of deadline. > > > > > > > > > > > > > > Best, > > > > > > > Jingsong > > > > > > > > > > > > > > On Mon, Oct 19, 2020 at 5:36 PM Kurt Young <ykt...@gmail.com> > > > wrote: > > > > > > > > > > > > > > > Can we change the freeze date to October 30th (Friday next > > > week)? It > > > > > would > > > > > > > > be helpful > > > > > > > > for us if we have 2 more days. > > > > > > > > > > > > > > > > Best, > > > > > > > > Kurt > > > > > > > > > > > > > > > > > > > > > > > > On Mon, Oct 19, 2020 at 5:00 PM Robert Metzger < > > > rmetz...@apache.org> > > > > > > > > wrote: > > > > > > > > > > > > > > > > > Hi all, > > > > > > > > > > > > > > > > > > Dian and I would like to discuss a few items regarding the > > > upcoming > > > > > Flink > > > > > > > > > 1.12 feature freeze: > > > > > > > > > > > > > > > > > > *A) Exact feature freeze day* > > > > > > > > > So far, we've always said "end of October > > > > > > > > > < > > > https://cwiki.apache.org/confluence/display/FLINK/1.12+Release>" for > > > > > > > > the > > > > > > > > > freeze. We propose (end of day CEST) October 28th > (Wednesday > > > next > > > > > week) > > > > > > > > as > > > > > > > > > the feature freeze time. > > > > > > > > > We want to create RC0 on the day after the feature freeze, > to > > > make > > > > > sure > > > > > > > > the > > > > > > > > > RC creation process is running smoothly, and to have a > common > > > testing > > > > > > > > > reference point. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *B) What does feature freeze mean?*After the feature > freeze, > > > no new > > > > > > > > > features are allowed to be merged to master. Only bug fixes > > and > > > > > > > > > documentation improvements. > > > > > > > > > The release managers will revert new feature commits after > > the > > > feature > > > > > > > > > freeze. > > > > > > > > > Rational: The goal of the feature freeze phase is to > improve > > > the > > > > > system > > > > > > > > > stability by addressing known bugs. New features tend to > > > introduce new > > > > > > > > > instabilities, which would prolong the release process. > > > > > > > > > If you need to merge a new feature after the freeze, please > > > open a > > > > > > > > > discussion on the dev@ list. If there are no objections > by a > > > PMC > > > > > member > > > > > > > > > within 48 (workday)hours, the feature can be merged. > > > > > > > > > > > > > > > > > > *C) When to cut the "release-1.12" branch off master?* > > > > > > > > > In the last feature freeze, we had a pretty lengthy phase > of > > > > > maintaining > > > > > > > > > the "master" and "release-1.11" branches with the same > fixes. > > > > > Therefore, > > > > > > > > I > > > > > > > > > would like to propose an adjustment to the release process: > > We > > > will > > > > > have > > > > > > > > a > > > > > > > > > stabilization phase on master, between the feature freeze > and > > > the > > > > > branch > > > > > > > > > cut. > > > > > > > > > I expect this stabilization phase to last between 1 and 3 > > > weeks, > > > > > > > > depending > > > > > > > > > on the issues we find. Once all blockers are resolved, and > no > > > new > > > > > > > > blockers > > > > > > > > > are surfacing, we can cut off the "release-1.12" branch and > > > finalize > > > > > the > > > > > > > > > release. > > > > > > > > > Is anybody in the community waiting for the cut off to > happen > > > sooner > > > > > so > > > > > > > > > that they can merge a big feature to Flink 1.13 ? (if that > > > would be > > > > > the > > > > > > > > > case, then we can not have a stabilization phase) > > > > > > > > > > > > > > > > > > > > > > > > > > > Let me know what you think! > > > > > > > > > > > > > > > > > > Best, > > > > > > > > > Dian and Robert > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > -- > > > > > > > Best, Jingsong Lee > > > > > > > > > > > > > > > > > > > > > > -- Best, Jingsong Lee
